Handled RF spectrum analyser using Arduino

(1)

Universidade de Aveiro Departamento deElectrónica, Telecomunicações e Informática 2013

Vitor Miguel

Melo Soares

Analisador de espectros portátil para RF usando

Arduino

(2)

(3)

Universidade de Aveiro Departamento deElectrónica, Telecomunicações e Informática 2013

Vitor Miguel

Melo Soares

Analisador de espectros portátil para RF usando

Arduino

Handled RF spectrum analyser using Arduino

Dissertação apresentada à Universidade de Aveiro para cumprimento dos requesitos necessários à obtenção do grau de Mestre em Engenharia Eletrónica e Telecomunicações, realizada sob a orientação científica do Pro-fessor Doutor Pedro Miguel Cabral, ProPro-fessor Auxiliar do Departamento de Eletrónica, Telecomunicações e Informática, da Universidade de Aveiro

(4)

(5)

o júri / the jury

presidente / president Professor Doutor Nuno Miguel Goncalves Borges de Carvalho Professor Catedrático da Universidade de Aveiro

vogais / examiners committee Professor Doutor Pedro Renato Tavares de Pinho

Professor Adjunto, Area Departamental de Engenharia de Eletrónica e Teleco-municações e de Computadores do Instituto Superior de Engenharia de Lisboa (Arguente)

Professor Doutor Pedro Miguel da Silva Cabral Professor Auxiliar da Universidade de Aveiro (Orientador)

(6)

(7)

agradecimentos / acknowledgements

Fazer acontecer reside na vontade prórpia de cada um, assim sendo, aproveito este espaço para expressar um pouco de emoção neste documento técnico. Naturalmente faço referência aos meus pais pela sua vontade de me possi-bilitar acesso a conhecimento. Aos meus colegas de estudo que se tornaram grandes amigos durante esta aventura, que da qual esta tese, assinala apenas uma etapa para nós.

Todas as pessoas que envolvidas directa ou indirectamente, possibilitaram a realização deste trabalho, como o meu prestável orientador Doutor Pe-dro Cabral, o hábil soldador e desenhista de placas para circuitos integrados Paulo Gonçalves ou o Sr. Pereira que nos aturou todos estes anos de curso e nós a ele.

É também normal referir que não é possível mencionar toda a gente com quem partilhamos apoio, mas para todos estes que serão recordados por mim, desejo muito amor, paz e dedicação nos altos e baixos desta vida energética, que são tão bem representados por uma onda sinusoidal.

(8)

(9)

Palavras Chave Analisador de espectros, RF, Rádio-frequência, Arduino

Resumo Rádio-frequência é o termo utilizado para designar a gama de frequêcias utilizadas na transmissão de sinais eletromagnéticos, através do meio livre. O domínio deste fenómeno possibilita a que dispositivos eletrónicos possam comunicar sem fios.

A análise do espectro eletromagnético utilizado pelos diferentes dispositivos que coabitam no dia a dia, é essencial para possibilitar o correcto funciona-mento de todos. Para tal, são utilizados analisadores de espectro, que com os quais se pode monitorizar algumas das características físicas das comuni-cações sem fios.

O objetivo deste trabalho é documentar o projeto, implementação e testes de validação de um analisador de espectros portátil para rádio-frequência utilizando tecnologia de controlo digital opensource, Arduino.

(10)

(11)

Key words Spectrum analyser, RF, Radio Frequency, Arduino

Abstract Radio frequency term refers to the electromagnetic spectrum bandwidth used to transmission of electromagnetic signals trough free space.

The human knowledge about this phenomena turns possible to set wireless communications between electronic devices.

The analysis of that spectrum is demanding in order to achieve proper com-munications between those devices. Whit spectrum analysers is possible to observe some physical characteristics of the communication.

The aim of this document is to furnish information about the project, design and implementation of a spectrum analyser using open source digital control technology, Arduino.

(12)

(13)

List of Figures

2.1 Radio communications environment . . . 3

2.2 Signal representation . . . 4

2.3 Ideal and real output spectrum representations . . . 5

2.4 Frequency domain and time domain measurements . . . 6

2.5 Frequency domain analysis of a two tone signal . . . 7

2.6 RF front-end receiver . . . 8

2.7 Super-heterodyne architecture . . . 9

2.8 Swept-tune principle applied to a CRT screen . . . 10

2.9 Simplified FFT topology . . . 10

2.10 Frequency range example of 0.7 GHz . . . 11

2.11 Accuracy of amplitude over frequency measurement . . . 12

2.12 Two indistinguishable input signals . . . 12

2.13 Input to output noise increase . . . 13

2.14 Sensitivity and dynamic range depiction . . . 14

2.15 Voltage Standing Wave Ratio . . . 15

2.16 1 dB compression point illustration . . . 16

2.17 Noise figure example . . . 17

2.18 Distortion spectrum . . . 18

2.19 Non linear interference inside the desired band . . . 18

2.20 IP3 and IMR examples . . . 19

3.1 Simplified Homodyne receiver . . . 22

3.2 Image frequency rejected with a LPF . . . 22

3.3 Image frequency problem in a heterodyne receiver . . . 23

3.4 Image rejection and channel select in heterodyne receivers . . . 24

3.5 Rejection of image and channel selection differences for a high or low IF . . . . 24

3.6 Spectrum analyser prototype block diagram . . . 25

3.7 LO sweep . . . 26

3.8 Gain block . . . 27

3.9 Mixer ports diagram . . . 28

3.10 IF filter bank . . . 30

4.1 Electric diagram of HMC374 evaluation board . . . 35

4.2 Low noise amplifier HMC374 evaluation board . . . 35

4.3 Low noise amplifier HMC374 gain and VSWR . . . 36

(18)

4.5 Amplifier HMC580ST89 evaluation board . . . 37

4.6 Power gain and VSWR from HMC580ST89 . . . 38

4.7 LO amplification stage . . . 38

4.8 Electric diagram of ERA-5+ evaluation board . . . 39

4.9 Amplifier ERA-5+ evaluation board . . . 39

4.10 Experimental gain and VSWR plots from ERA-5+ . . . 40

4.11 Electric diagram of HMC207AS8 evaluation board . . . 40

4.12 Mixer HMC270AS8 . . . 41

4.13 Experimental conversion loss . . . 41

4.14 Mixer HMC270AS8 conversion loss . . . 41

4.15 Electric diagram of VCO CVCO55BE-0800-1600 evaluation board . . . 42

4.16 VCO CVCO55BE-0800-1600 evaluation board . . . 42

4.17 Electric diagram BPF BP60370 . . . 43

4.18 BPF BP60370 insertion loss and VSWR . . . 44

4.23 Filters evaluation board . . . 46

4.24 Electric diagram RF switch HMC252QS24 . . . 47

4.25 RF switch HMC252QS2 evaluation board . . . 48

4.26 RF switch HMC252QS2 insertion loss and Voltage Standing Wave Ratio (VSWR) 48 4.27 Power combiner ZSC-4-1+ . . . 49

4.28 Power splitter ZSC-4-1+ insertion loss and VSWR . . . 50

4.29 Electric diagram of LTC5507 evaluation board . . . 51

4.30 Power detector LTC5507 evaluation board . . . 51

4.31 LTC5507 Experimental output curves . . . 51

4.32 Electric diagram - Arduino . . . 53

4.33 PWM wave form to produce a saw tooth . . . 54

4.34 Keyboard design . . . 55

4.35 Screen coordinate system for pixels . . . 55

4.36 Interface board . . . 56

5.1 Prototype test setup . . . 57

5.2 RF chain overview . . . 59

5.3 Software flowchart . . . 60

5.4 Interface menu . . . 60

5.5 VCO output frequency over duty-cycle control . . . 61

5.6 Linear interpolation results . . . 62

5.7 Amplitude over frequency addressed by duty-cycle manually controlled . . . 63

5.8 Practical measurements, 1 tone . . . 65

5.9 Practical measurements, 2 tone . . . 66

B.1 Two port network diagram . . . 71

B.2 Two port network diagram with S-Parameters representation . . . 72

(19)

D.1 Power gain and VSWR from HMC580ST89 . . . 74

E.1 RF switch HMC252QS24 insertion loss and VSWR . . . 75

F.1 Power combiner insertion loss and VSWR for channel 2 and 3 . . . 76

H.1 Practical measurements . . . 78

I.1 LNA HMC374 PCB schematic . . . 79

I.2 Amplifier HMC580ST89 PCB schematic . . . 79

I.3 Mixer HMC207AS8 PCB schematic . . . 79

I.4 VCO CVCO55BE-0800-1600 PCB schematic . . . 80

I.5 Filter bank PCB schematic . . . 80

(20)

(21)

List of Tables

2.1 Output components resulting from a two tone test . . . 19

4.1 LNA HMC374 specifications . . . 35

4.2 General amplifier HMC580ST89 specifications . . . 37

4.3 General amplifier ERA5+ specifications . . . 39

4.4 Mixer HMC207AS8 specifications . . . 41

4.5 Voltage controlled oscillator CVCO55BE-0800-1600 specifications . . . 42

4.6 BPF BP60370 . . . 43

4.7 BPF BP60290 . . . 44

4.8 BPF BP60110 . . . 45

4.9 RF switch HMC252QS24 specifications . . . 47

4.10 RF switch HMC252QS24 truth table . . . 47

4.11 Power splitter ZSC-4-1+ . . . 49

4.12 Power detector LTC5507 specifications . . . 50

4.13 Arduino Uno R₃ board specifications . . . 52

4.14 Color TFT Shield specifications . . . 55

(22)

(23)

Acronyms

ADC Analog to Digital Converter ADNL Average Displayed Noise Level

ANACOM Autoridade Nacional de Comunicações BGA Ball Grid Array

BPF Band Pass Filter CRT Cathode Ray Tube

DAC Digital to Analogue Converter DC Direct Current

DFT Discrete Fourier Transform DSP Digital Signal Processor DUT Device Under Test

ENIG Electroless Nickel/Immersion Gold

EEPROM Electrically Erasable Programmable Read Only Memory FFT Fast Fourier Transform

FM Frequency Modulated

FPGA Field Programmable Gate Array

GSM Global System for Mobile Communications HASL Hot Air Solder Leveling

HPF High Pass Filter IC Integrated Circuit

IDE Integrated Development Environment IF Intermediate Frequency

(24)

IMR Intermodulation Ratio IO Input Output

IP3 Third-order Intercept Point IT Instituto de Telecomunicações

ITIS Immersion Tin and Immersion Silver LCD Liquid Crystal Display

LNA Low Noise Amplifier LO Local Oscillator LPF Low Pass Filter

MDS Minimum Detectable Signal NF Noise Figure

OCXO Oven Controlled Crystal Oscillator OSP Organic Solder Preservative

P1dB 1 dB Compression Point PAE Power Added Efficiency PCB Printed Circuit Board PWM Pulse Width Modulation QPSK Quadrature Phase-Shift Keying RAM Random Access Memory

RBW Resolution Bandwidth RF Radio Frequency

RTSA Real Time Spectrum Analyser SAW Surface Acoustic Wave

SD Secure Digital

SDR Software Defined Radio SMA SubMiniature Version A SNR Signal to Noise Ratio SPI Serial Peripheral Interface

(25)

TCXO Temperature Compensated Crystal Oscillator TFT Thin-Film-Transistor

UART Universal Asynchronous Receiver Transmitter USB Universal Serial Bus

VCO Voltage Controlled Oscillator VSWR Voltage Standing Wave Ratio VSA Vector Signal Analyser

(26)

(27)

Chapter 1

Introduction

1.1 Motivation

Wireless technology became a reality in our daily life. Electronic devices are able to exchange data through atmosphere and spreading for many different proposes. It may be transparent to human senses though there are physical constraints on data transmission over air and the comprehension of what is happening is key to turn all that possible. As for example, each established communication link between a transmitter and a receiver requires a bandwidth portion of electromagnetic spectrum.

Moreover the proliferation of radio broadcasts, cellular networks and domestic wireless applications would be interfering with each other if there were no restrictions on the electro-magnetic spectrum usage. As the communications take place in a shared medium it is essential they can cohabit.

The state of the art of Radio Frequency (RF) technology restricted to such physical lim-itations, keeps looking to improve communications in a crowded spectrum. Engineers need to understand the conditions where wireless links are done and find solutions to come up to actual demanding standards. In practice this means to manipulate physical characteristics of electromagnetic signals to achieve transmission efficiency. There are necessary measure-ment tools to go ahead with telecommunications and one of them is denominated as spectrum analyser.

Since the begin of radio communications, spectrum analysers are used as an elementary instrument to inspect the used spectrum. Till today, they are a valued instrument for RF research and development laboratories, private companies operating in telecommunications business and governmental organizations which regulate wireless communications in open air. Due to the quest for wireless communications, these instruments are needed to operate in indoor and outdoor environments. Benchtop versions may afford powerful processors and many other components for superior performance, turning to be heavier and bigger. Handled versions that can run on batteries, are developed to fill the need to identify interfering RF sources on field that may corrupt other wireless systems. It makes them useful tools to improve and troubleshoot wireless networks.

Any commercial version can be quite expensive. Describing the design and implementation of such instrument on a academic thesis, applying for open source software, may be a revision to the ones who need to access this kind of instruments.

(28)

1.2 Objectives

The aim of this dissertation is to project and implement a handled spectrum analyser prototype. The device should be capable to perform power measurements over frequency in a selectable bandwidth range from 880 to 950 MHz.

It is pretended to pick a Resolution Bandwidth (RBW) channel amongst different ones. Other parameters like sweeping times and measurement tools should be available to the user. The utilization of these devices is done trough a graphic display and a keypad which also are contemplated in this document.

It will be described the necessary concepts to understand the hardware design, software control directives and practical tests to implement such device.

1.3 Dissertation outline

This document is formulated in 6 chapters and appendices, conducting the reader through the necessary subjects to project and understand a spectrum analyser.

Chapter 1 is this introduction where motivation and objectives are detailed

Background theory is introduced in chapter 2, elucidating the tackled topics to comprehend what is spectrum analysis and how to do it.

To choose an appropriate architecture in chapter 3 different approaches about collect radio signals are briefed.

The different components used to implement the spectrum analyser prototype are detailed in chapter 4.

The tests and results about the circuit functionality are in chapter 5 Concluding this document is chapter 6 with final considerations.

In appendices are additional informations to expand description details about radio circuits validation, tests results, the used PCB designs, calculus and the Arduino source code.

(29)

Chapter 2

Background theory

In telecommunications there is the objective to transmit data from point a to point b. When the channel between the two points is the atmosphere the transmitter and the receiver are named as radios.

The process consists to convert the intended data (audio, video, computer data, etc.) into an electromagnetic signal with proper characteristics to spread trough open space. This involves signal conditioning, modulation and frequency conversion. All these processes have physical requirements which often take place in harsh environments.

This requires very well optimized transmission techniques and spectral optimization to transmit maximum data with least bandwidth1_.

Spectrum analysers are a used tool to measure frequency utilization in communications. It is an essential piece to comprehend physical aspects about radio communications where the human senses are not able to perceive it.

Describing what is a signal, frequency spectrum, time and frequency domain measure-ments, operation theory of a spectrum analyser and main specifications of such instrument will introduce the reader to what is involved with spectra analysis.

Figure 2.1: Radio communications environment

1

There is a maximum data transmission limit over a fixed bandwidth which calculus are attributed to Claude Shannon [1].

(30)

2.1 Signals

Signals are temporal functions that may provide useful information. Traditionally repre-sent physical characteristics of processes and can be either natural or synthesized. Furthermore they are modulated to encode data.

In electric systems a signal is a voltage or current variation [2]. It can be graphically represented as in figure 2.2. The signal present in 2.2a might be not predictable. It is not easy to mathematically describe it.

With the propose to evaluate signal transmission over networks, electronic science use controlled sinusoidal signals as depicted in figure 2.2b. These sinusoidal functions are mathe-matically represented by an the amplitude A (in volts) and period T (in seconds). A sinusoid signal may be mathematically expressed as:

v(t) = A sin(ωt + Φ) (2.1)

The angular frequency of a sinusoid is represented with ω = 2πf (rad/s) and period is defined as the inverse of frequency stating f = _T1Hz. Φ represents the signal phase. It can be understood as a time shift in electromagnetic signals, but for now it will not be take in consideration.

More complex signals can be represented as a sum of diverse sinusoids. Equation 2.2 expresses the stated before.

v(t) = A0sin(ω0t) + A1sin(ω1t) + ... + Ansin(ωnt) (2.2)

This signal description resulting from the sum of sinusoids was introduced by Jean Fourier. Fourier series and Fourier transform are mathematical tools widely used in signal character-ization and processing. They have an important role in spectra analysis once they link time domain with frequency domain [2, 3, 4].

(a) Arbitrary signal (b) Sinusoidal signal Figure 2.2: Signal representation

(31)

2.2 Frequency spectrum

As mentioned in equation 2.2 it is possible to perceive that a signal can contain n different frequency components represented as ωn or fn. This concept allows to modulate signals and

is how telecommunications techniques controls and processes signals [5]. Considering three independent signals v0(t), v1(t) and v2(t)

v0(t) = sin(2πf0t), v1(t) =

1

3sin(2πf1t), v2(t) = 1

5sin(2πf2t) (2.3) The sum of these signals leads to a new one, represented as v(t):

v(t) = v0(t) + v1(t) + v2(t) (2.4)

Assuming f₀ = 1f , f1 = 3f and f2 = 5f it is possible now to say that signal v(t) will have

three different frequency components and this represents the signal’s spectrum. This example relates to the first three frequency components which characterize a square wave.

Ideally the output spectrum for a single frequency oscillator may be represented as in figure 2.3a. Because of noise both frequency and amplitude are affected, turning the real world spectrum components to be seen with skirts as represented in figure.

Concepts like spectral occupancy start to emerge. In the case of signal v(t), it requires 5f Hz of bandwidth. As an example about the practical importance of this, adjacent com-munication links which operates at different frequency bands need to restrain their spectral emissions. Otherwise they can interfere with the surrounding channels. This situation it is not desired because it may corrupt transmitted data.

As a practical example of spectrum usage and regulation in Portugal, Autoridade Nacional de Comunicações (ANACOM) which is the governmental agency to legal conduct communi-cations, sets the frequency band from 87,5 MHz to 108 MHz for Frequency Modulated (FM) radio broadcasting [6]. Another example is the regulated bandwidth for Global System for Mobile Communications (GSM) which fits in the 880 to 890 MHz band plus 925 to 930 MHz, which the intended prototype for this dissertation will focus.

(a) Ideal oscillator (b) Real oscillator Figure 2.3: Ideal and real spectrum representations [7]

(32)

2.3 Time and frequency domain

In figure 2.4 is represented a three-dimensional coordinate system that will be used to understood signal characteristics. In the axes are represented amplitude, frequency and time. Signals v₀(t), v1(t), v2(t) and v(t) are also pictured.

With this perspective it is possible to analyse signals into two distinct domains, in time or frequency. Selecting one view leads to different measurements about them [3, 8].

While signal v(t) is mathematically described in equation 2.4 and it may be produced by a signal generator. It is possible to analyse it in time domain using an oscilloscope2and a square wave approximation will be perceived (figure 2.4). In this case is feasible to characterize the signal in terms of amplitude and period.

Some details about a signal can be only known in time domain as pulse rise or fall times, overshoot and ringing [3]. Anyhow the three present signals which describe v(t) are indistin-guishable in a time domain analysis, yet they are in the frequency domain.

Spectrum analysers take place when it is necessary to analyse frequency content. These tools allow to analyse the signal composition and how much power exists over specific fre-quencies. More than that it turns to be a better approach to understand harmonic content or analyse the occupied bandwidth for a communication channel.

As an example of a frequency domain measurement, figure 2.5 represents a spectrum analyser view of a two tone signal3_.

Figure 2.4: Frequency domain and time domain measurements

2

An oscilloscope is a device used to perform electronic measurements. It allows to perceive voltage variation over time.

3

A two tone signal is a test signal used in telecommunications. The main characteristics of particular signal, are two distinct spectral components. More about this signal will be detailed in later sections.

(33)

Figure 2.5: Frequency domain analysis of a two tone signal

2.4 Spectrum analyser theory of operation

To perform spectral analysis there are few distinct techniques. The essential function of these devices is to display power over frequency, though additional measurements can be performed pursuant the used method.

A spectrum analysis may be done under a swept-tune method or as a result of math calculations after the signal being acquired in time. Both provide the display of amplitude versus frequency.

Swept-tune methods interpret the power over frequency under successive tuning scans, while the other approaches use Fast Fourier Transform (FFT) techniques after an Analog to Digital Converter (ADC) had sampled in time the signal provided by a radio front-end.

For the new requirements of RF transmission techniques, where it can be mentioned com-plex digital modulations or spread spectrum techniques, spectrum analysis is commended to state of the art analysers where they can be commonly find with the name of signal analyser. These are the modern measurement equipments that can provide a more exhaustive signal analysis [9, 10].

2.4.1 Radio receiver

Before further details, any spectrum analyser has a RF receiver front-end. This hardware piece is essential for conditioning the electromagnetic signals and select the desired frequency band to process [11, 12, 13].

A RF front-end may be defined as anything between the antenna and the Intermediate Frequency (IF) stage [14]. It can be represent as plainly as figure 2.6.

The collected signals by the antenna are delivered to a variable attenuator to prevent high power levels to reach more sensible components. A Band Pass Filter (BPF) or a Low Pass Filter (LPF) selects the RF band, delivering the signal band to the respective interpretation process.

This simplicity might be not so efficient when treating RF or microwave frequency signals or for low power levels. Signal detection may be impractical even for sophisticated ADCs.

Turning a RF front-end as simple as possible is still object of study and research, especially for the new radio generation, the Software Defined Radio (SDR) [10].

(34)

Radio front-ends are also projected to provide frequency translation to a lower frequency level, easy to process. Special amplifiers with low noise characteristics are conjugated to improve powerless signals detection.

Concerning about interpret power over frequency, swept-tune methods have given proofs of functionality over the years and are widely used in spectrum analysers and many other wireless devices. Alongside there are the digital denominated spectrum analysers providing flexibility following up RF research needs.

A description of both methods will be presented next.

Figure 2.6: RF front-end receiver

2.4.2 Swept-tune method

Swept-tune spectrum analysers uses super-heterodyne radio architecture to sweep the input frequencies and display the energy present at each tune sweep step [13]. Super-heterodyne is a RF receiver architecture characterized by frequency conversion and its tuning ability. Figure 2.7 describes a generic super-heterodyne architecture.

A Low Noise Amplifier (LNA) receives the incoming RF signal amplifying the weak signals in the frequency band of interest to higher levels.

The mixer in the circuit is responsible to convert RF frequency to a IF one, with a con-trolled Local Oscillator (LO). This brings to the receivers tuning ability once varying the LO frequency, the output of the mixer will be directly related to the desired input frequency.

That conversion is intended to a fixed frequency for which the BPF is centred tuned. This rejects other conversion terms off the mixer selecting just the wanted channel. This is also known as IF filtering. The signal is again given into an amplifier to recover from previous attenuations and is ready to be interpreted.

Earliest spectrum analysers at the output of the IF, were plugged to a respective con-ditioning circuit to display amplitude over frequency in a Cathode Ray Tube (CRT) screen [3, 4, 13]. Figure 2.8 illustrates that IF interpretation section for swept-tune analysers.

A ramp generator guides the LO sweeping frequency as the same time promotes the electron beam deflection in the horizontal axis. The amplitude from the IF frequency caught in the envelope detector deflects the vertical axis. As result amplitude is displayed over to the frequency.

Whit the aid of digital era, microprocessors and other digital devices, turns possible other ways to interpret the IF frequency. This thesis project will have focus in another alternatives to those analogue components and these will be discussed later.

(35)

A relevant disadvantage relatively to other spectral analysis methods is the inability to measure the signal phase. Also this method may miss some quick events as pulses which skips to detection while LO is sweeping the input.

Figure 2.7: Super-heterodyne architecture

2.4.3 Fast Fourier Transform topology

This kind of spectrum analysers are also known as digital spectrum analysers in virtue to the calculation spectrum process. Spectrum calculations are sustained by Fourier transforms which involves serious mathematical theory. The most notable algorithm that converts a series of time domain samples into their spectral components is known as Fast Fourier Transform (FFT) which derives from the Discrete Fourier Transform (DFT) [15, 16, 17].

Figure 2.9 shows a block diagram with the signal acquisition principle of operation. The signal is collected by an ADC which samples and quantizes it in time domain. Furthermore powerful computing capability is required and devices as Digital Signal Processors (DSPs) or Field Programmable Gate Arrays (FPGAs) are often used.

Down-conversion, filtering and amplitude detection may be digitally computed. These operations are implemented with FFTs tools, converting the collected signal data to frequency components [4, 13, 18].

With this approach it is possible to retrieve phase information which brings a deeper knowledge about the signal. Though it has major limitations in the input frequency bandwidth which is limited to the ADC sampling rate. High sampling speeds can be obtained with less resolution bits, but this will restrict resolution, accuracy and dynamic range [19].

Vector signal and real-time analysers

More demanding signal analysis is required to detect modulation parameters and transient events. Engineers had to improve and adapt the analysis techniques in order to the dynamic RF panorama. Both Vector Signal Analyser (VSA) and Real Time Spectrum Analyser (RTSA) came to satisfy the new needs. Furthermore these models use Fourier math tools alongside

(36)

with sweeping techniques and other technologies working all together to enhance spectrum analyse.

VSAs are able to perform measurements over modulation parameters. Also can display power versus time and spectrograms4. Specifications and features as the different perceived modulation types change from model to model. These exemplars are not indicated to perform real-time analysis once the signal needs to be in memory before calculations and the signal sampling process may not be done constantly [13, 19].

In RTSA models there is a real-time processing stage before signal post processing, grant-ing the detection of quick changes in the input signal. Digital phosphor screens are used to implement functionalities as trace persistence and proportionality which are not natural features to digitally controlled Liquid Crystal Displays (LCDs) [20].

Measurements which requires to be digital processed as demodulation, are post executed and signal acquisition can be continuous. This conceives a spectral analysis coherent with the variations of the input signal.

Figure 2.8: Swept-tune principle applied to a CRT screen [3]

Figure 2.9: Simplified FFT topology [4, 13]

4

A spectrogram is a color graphic representation of a spectrum. It displays frequency variation over time while color intensity changes to represent power.

(37)

2.5 Spectrum analyser characteristics

This section explains the main characteristics of a spectrum analyser. These may be designated as figures of merit and are generally known as the device specifications. They are straight related with physical circuit characteristics and limitations. This is what distinguish one equipment from another and serve as guide selection to the user needs.

Any spectrum analyser has specifications as frequency range, accuracy, resolution, sensi-tivity, dynamic range and distortion parameters [3, 8, 10, 21].

2.5.1 Frequency range

This specification tells the minimum and maximum frequencies which the device is able to analyse. The difference of maximum and minimum tells the bandwidth of operation. It is also called as the span, and the term full span is used when it is performed a analysis over the full range.

This characteristic is dictated by the RF front-end, where the frequency responses of the antenna, LPF, LNA, the mixer and the LO ranges all combined together define the frequency range.

Figure 2.10: Frequency range example of 0.7 GHz

2.5.2 Accuracy

In a measurement context this figure of merit represents how close is the performed mea-surement to the true present value [22]. A device tend to be more accurate as closer to the original value it can read the input data.

Spectrum analysers specify different types of accuracy and they are listed as frequency, amplitude and phase. Figure 2.11 illustrates the lack of accuracy of a amplitude over frequency a measurement.

Note: It came in context to distinguish accuracy from precision. The last one is related to the capability to reproduce the same readout for the same input value.

Frequency accuracy

This figure of merit mainly relies on the accuracy of the local oscillator. Frequency refer-ences are subject of fluctuations due to temperature variations. For this reason manufacturers tend to implement internal frequency references with Temperature Compensated Crystal Os-cillator (TCXO) and Oven Controlled Crystal OsOs-cillator (OCXO) in spectrum analysers [4].

(38)

Figure 2.11: Accuracy of amplitude over frequency measurement

Amplitude accuracy

To correct identify the power level at the input of the spectrum analyser it is need to take in account the accuracy of all components in the circuit. All the losses and gains need to be accounted for subsequently correct the readout value.

2.5.3 Resolution

Frequency RBW5 describes the spectrum analyser competence to distinguish adjacent frequency components. When at least two spectral elements are less spaced than the frequency resolution, they will be indistinguishable in the screen. Generally manufacturers provide different resolution bandwidths to satisfy distinctive needs.

Figure 2.12 shows a filter shape responsible for the RBW. Inside the BPF bandwidth exists two signals that wont be solved on the display. Narrower bandwidth filters will permit to distinguish closer signals.

Figure 2.12: Two indistinguishable input signals

(39)

2.5.4 Sensitivity

Any radio receiver has a noise floor level and therefore any signal with an inferior power level will be unrecognisable. Sensitivity informs the Minimum Detectable Signal (MDS) above the Average Displayed Noise Level (ADNL).

Noise floor is originated by thermal noise, inner components noise and is proportional to the bandwidth. Inspect wider bandwidths increasing RBW commit sensitivity.

Figure 2.13: Input to output noise increase

Considering signal power Pi and noise Ni at input terminal. The output noise level is

related in equation 2.5. This establish the receiver noise floor in Watts.

N0= k · T0· F · B (2.5)

Where k is the Boltzman constant(k = 1.38 × 10−23J/K), T0 denotes the temperature

in kelvin (290 K as room temperature), N F is the noise figure and all th is related to the operating bandwidth B in Hz.

Expressing equation 2.5 in dBm it comes:

N0(dBm) = −174(dBm/Hz) + N F (dB) + 10log10(B) (2.6)

A detectable signal needs to be greater than that. Combining the MDS with a minimum desired Signal to Noise Ratio (SNR) bring the figure of merit sensitivity[24]. It is usual to consider a minimum 3 dB stronger than noise floor [25].

Si= −174 + N F (db) + 10log10(B) + SN Rmin (2.7)

One characteristic of the MDS is the SNR which have to be high enough. This is depicted in figure 2.14.

2.5.5 Dynamic range

The dynamic range specifies the spectrum analyser capability to handle with weak and strong signals at the same time. It consists on a ratio and is expressed in dB. Figure 2.14 graphically illustrates the dynamic range relation in a measurement.

Maximum detectable signal is limited to components saturation levels while minimum signal detection is limited by the instrument noise floor and contrived signals. The last quoted problem are the second and third order intermodulation distortion products which may affect dynamic range [23].

2.5.6 Distortion

Spectrum analysers may be wanted to realize distortion measurements. For this reason it is necessary to prior now inner distortion levels. This is related to the internal spectrum

(40)

Figure 2.14: Sensitivity and dynamic range depiction

analyser components distortion. It needs to be take in account once measuring others devices distortion.

Distortion is an undesired but unavoidable effect of the non ideal components functionality. Moreover Intermodulation Distortion (IMD) is of great importance and will be detailed later together Third-order Intercept Point (IP3).

2.6 RF design basic concepts

In a RF design project there are useful concepts which helps understanding circuit blocks characteristics while performing. This notions are important to the component evaluation process. Some of them may have impact in the circuit functionality leading to respective spectrum analyser specifications. Special attention to RF electronic parts is demanding once high operating frequencies require some specific evaluation methods to ensure maximum power transfer.

Decibels, decibels to milliwatt and S-parameters are used to characterize a input-output device related to the enclosed power levels, are described in appendix A and B for whom are not familiarized with it.

This section provides an explanation about VSWR, 1 dB compression point, noise figure and intermodulation effects.

2.6.1 Voltage Standing Wave Ratio

In RF circuits, input and output VSWR tells how well the power transfers occurs for a frequency range with specific impedance value. It provides a way to analyse the impact of reflections in signal transmission, telling the ratio of the maximum to minimum values that voltage and current can ever get [26, 27].

When two waves with the same frequency travelling in same medium but in opposite directions they sum each other resulting into a composite standing wave6 [26, 28, 29, 30].

In electric circuits either bad junctions or impedance variations over the network are re-sponsible for signal reflections. Connecting one component to another results in some of the incident wave to be reflected. This leads to the two waves propagating in opposite directions.

6

A pure standing wave forms only when the incident and reflected wave have the same amplitude and frequency[31].

(41)

As they sum each other occurs maximum constructive peaks and a minimums when both waves are in phase or opposition respectively (figure 2.15).

It is mathematically expressed in in equation 2.8, in order to the reflection coefficient provided by S11.

V SW R = 1 + |S11| 1 − |S11|

(2.8) An ideal VSWR of 1 would mean that all energy is being transmitted as result of no reflections S₁₁ = 0. This situation could be exemplified as an perfect impedance match. In case of a full signal reflection S₁₁= 1, would result on an infinite VSWR.

Figure 2.15: Voltage Standing Wave Ratio

2.6.2 Gain compression

Saturation effect in RF circuits is commonly designated as 1 dB Compression Point (P1dB). While systems are designed to operate in a linear region, they may change the desired be-haviour under certain circumstances.

For example, an amplifier output power will increase 1 dB for each dB increment in the input signal. The device will operate in this linear region till the input power reach such high value where the output power stops to increase in the same 1 dB proportion as it did before. Figure 2.16 illustrates an amplifier’s 10 dB gain curve. The linear behaviour starts do change when input power reaches almost 10 dBm. In this example, when input power increases to 11 dBm the output remains on 20 dBm not keeping the 1 dB gain proportion.

This depicts the P1dB which is defined as the output power level where the the gain curve is 1 dB compressed in relation to the linear region. In this example the output power level for P1dB is 20 dBm.

Besides amplifiers, another component type that can be defined in terms of compression is the mixer which in contrast with amplifiers, P1dB tend to be defined as the maximum input level for each mixer starts to produce a non linear output curve.

2.6.3 Noise figure

As a figure of merit Noise Figure (NF) is a measurement to characterize the whole RF circuit and individual components with at least two ports as an input and output. Knowing

(42)

Figure 2.16: 1 dB compression point illustration

NF from individual circuits allows to calculate receptor sensitivity [32]. This figure of merit quantify the SNR degradation from the input to the output in a electric device.

Any component subjected to a temperature above zero kelvin superimpose inner noise as for example, the thermal noise in a resistor. That noise is added to a signal passing through the component with no added power gain to the signal7_.

In figure 2.17 is represented the input and output spectrum at respective amplifier’s ports. In 2.17a it is perceptible a ground noise of -80 dBm and a major frequency component at 1 GHz with -20 dBm. Figure 2.17b depicts the output with the 1 GHz component pushed to 0 dBm and the noise floor raised to -30 dBm.

It is noticeable a 20 dB gain at 1 GHz although the ground noise level has been amplified 30 dB. This noise floor increase is related to the intrinsic device noise, being represented as the noise figure and in this example N F = 10dB. As already mentioned it represents a degeneracy in SNR [7, 32], and is mathematically expressed as in equation 2.9. The term NF is the representation in dB of the last relation which is known as noise factor (F ).

F = SN Rinput SN Routput (2.9) N F = 10log(F ) (2.10) Ft= F1+ F2− 1 G1 +F3− 1 G1G2 + ... + Fn− 1 G1G2...Gn − 1 (2.11)

As a RF receiver consists in a few cascaded devices, noise figure allows to to calculate the overall noise figure. Following equation 2.11 considers a series of components with noise factor Fn and respective numerical gains Gn which calculation method can be find in [33]. As this

is intended to calculate the total system NF to find out sensitivity of the receiver, the first term is the most important [1, 10, 31]. Following terms are successively smaller due to the denominator increasing factor contributing less for the overall N F_t.

7

In a noiseless system there are no degradation in SNR. Signal power and intrinsic noise are attenuated or amplified for the same factor [7]

(43)

(a) Input spectrum (b) Output spectrum Figure 2.17: Noise figure example [32]

2.6.4 Third order intercept point

This figure of merit, Third-order Intercept Point remarks relevant non linear characteristics in RF circuits. Something important to mention about non linear characteristics for a two port device is the creation of new frequency components at the output that are not present in input signal [1]. In addition any component has non linear characteristics even though they are designed to operate under a linear region. Special attention is taken into amplifiers and mixers.

Two tone test is a common way to measure some of the non linear characteristics. It allows to measure the system bandwidth impact[1].

This test consists in two different frequency components with the same amplitude used as input signal into the Device Under Test (DUT). It is expressed in equation 2.12.

vi(t) = Acos(2πf1t) + Acos(2πf2t) (2.12)

vo(t) = a1vi(t) + a2vi2(t) + a3vi3(t) (2.13)

The output result may be represented mathematically in a power series as in equation 2.13. The two tone signal test is applied and at the output will appear a series of frequency components. The most important new components are represented in figure 2.18 and detailed in table 2.1 [1, 10].

One of the major problems of this kind of distortion is the inability to remove the new frequency components that fall into bandwidth of interest, very close to f1 and f2. In this case

a BPF permit to attenuate the higher order harmonic components, but barely affects 2f₁− f₂ neither 2f2− f1.

In figure 2.20a it shown the output power relation of the fundamental frequency f1applied

to the DUT and a third order component 2f₁− f₂ demonstrating the IP3 point. In opposition to a 1 dB/dB slope of the fundamental frequency, the third order component raises with a 3 dB/dB slope. Theoretically if a straight line is plotted long side to the gain curves, they will hypothetically intersect8 That intersection point is known as IP3.

8

This intersection does not occur in practice once both output curves start to compress due to higher order terms [10].

(44)

Figure 2.18: Distortion spectrum

Figure 2.19: Non linear interference inside the desired band

This analysis allows to evaluate the Intermodulation Ratio (IMR) which is connected with intermodulation distortion. It describes the ratio of the fundamental output within IMD new components. IM R = Pf1,o P2f1−f2,o (2.14) = Pf2,o P2f2−f1,o (2.15) Equation 2.14 and 2.15 can be expressed in dB units as follow:

IM R(dB) = Pf1,o(dBm) − P2f1−f2,o(dBm) (2.16)

= Pf2,o(dBm) − P2f2−f1,o(dBm) (2.17)

Figure 2.20b represents a two tone test performed in a spectrum analyser. Using the spectrum analyser measurements it is possible to calculate IP3 with the following expression:

IP 3(dBm) = Pf1,o(dBm) +

IM R(dB)

(45)

(a) Third order intercept point (b) IMR measure in a spectrum analyzer Figure 2.20: IP3 and IMR examples

Practical meaning _v Frequency components

Component resulting from f₁− f₁ and f₂− f₂ Direct Current (DC)

f1 fundamental frequency f1 f2 fundamental frequency f2 f1 second harmonic 2f1 f2 second harmonic 2f2 f1 third harmonic 3f1 f2 third harmonic 3f2

Second order intermodulation products f2± f1

Third order intermodulation products 2f1± f2

Third order intermodulation products 2f2± f1

(46)

(47)

Chapter 3

Architectures

Subsequently to the introductory concepts involved with spectral analysis and circuit de-sign, it is necessary to set the approach to implement the prototype. A spectrum analyser is mostly a radio receiver, which consists on a electric circuit. Even with the continuous de-velopment in the state of the art of Integrated Circuit (IC), aspects as cost, complexity, size number of external components and power dissipation are taken in preponderation to the final prototype solution[7].

The circuit control may be performed with digital technology. This brings advantages to the development stage due to flexibility to circuit control and human to interface.

Previously in subsection 2.4.2 was introduced one of the most commonly known spec-trum analysers architectures. That was necessary to describe the theory of operation of such instrument. Regardless there are other architectures to extract to analysis, high frequency electromagnetic signals and a quick overview will be driven.

Therefore in this chapter will be detailed the chosen architecture for the radio circuit and the impact of the components in the overall system. Alongside is described the digital interface to control circuit under human direction .

3.1 Receiver architectures

A radio architecture can be understood as the design rules to organize the electronic components in order to perform a logical set of functions [34]. This piece of hardware is intended to collect the electromagnetic signals, reject the unwanted frequency bands, amplify small signals and all of this with minimum of distortion and added noise. As result the received electric signal will be conditioned and primed for interpretation, which usually includes at last human understanding.

A very important functionality in a receiver is the frequency down conversion. This takes place on the mixer which main function is to multiply the incoming RF signal with the LO controlled frequency. This creates two new frequency components at the output of the mixer, the sum and difference [23]. As result in the IF stage will be f_IF which is mathematically expressed as f_IF = fRF ± fLO.

After the incoming signal passes through the mixer, the modulation present at fRF is

trans-lated to the new f_IF components. This brings the problem of image and different architectures deal with it in a proper way [7].

(48)

3.1.1 Homodyne receivers

This architecture likewise known as direct conversion or zero-IF, is conceived to translate the input frequency of interest to a DC center. The frequency of the local oscillator matches the frequency of interest, f_LO = fRF. Such method requires a high precision LO reference to

stay in tune with the pretended input. As result of the intended down conversion it is possible to predict in the IF stage two spectral components, fIF0 = 0 and fIF1 = 2fRF.

It is affordable to reject the unwanted component fIF = 2fRF with a LPF for this situation.

Figure 3.2 shows the two frequency components presents at the output of the mixer.

Some modulation techniques may not operate with this architecture1. A few modifications are available in the block diagram of figure 3.1 to solve this issue, though in a basic spectrum analyser there is no intention to interpret the modulated data.

More to say about some issues that occur in this architecture as DC offsets and power leakage from RF to LO and vice versa. This can be troublesome for signals integrity [7, 10, 23]. Direct conversion has wide use for single tuned frequency channels. The simplicity of the architecture avoid image problems, and provides effectiveness extracting audio signals straight forward from fIF [23].

Figure 3.1: Simplified Homodyne receiver

Figure 3.2: Image frequency rejected with a LPF

1

Signal modulation like FM or Quadrature Phase-Shift Keying (QPSK) convey information that would be lost with this simplified process. [7].

(49)

3.1.2 Heterodyne receivers

In this architecture2 the LO frequency does not equate the desired input frequency band. The new frequency conversion is intended for for other components that are not DC centred. Again, at the output of the mixer there are fIF = fRF− fLO and fIF = fRF+ fLO. Providing

a fairly higher centred fIF than in the homodyne architecture, allows to escape from obstacles

previously described, although problem of image demands for solutions.

Choosing IF center frequency and LO tuning frequency takes some aspects in considera-tion. Incoming frequencies centred at f_RF + 2fIF will fall inside the IF filtering stage being

denominated as image frequencies. Figure 3.3 illustrates the LPF output spectrum. The two incoming frequency components fRF and fimg equally distant from fLO fall inside the filter’s

pass band and that may invalidate the communication channel.

Figure 3.3: Image frequency problem in a heterodyne receiver

Image rejection and channel selection are conceivable with filters. Filtering in the RF stage removes the undesired images to reach the mixer. Anyhow, the mixer will produce both fIF0 and fIF1 and other spurious. A channel selection is performed with a LPF or BPF This

selects the wanted conversion term to establish communication channel.

In the architecture project is necessary to decide which high or low fIF side will be used to

establish the communication channel. Some trade off consequences happens as for example, the added complexity to achieve good filters centred for high frequencies causing a loss on resolution when comparing with filters projected to operate at lower frequencies.

A low f_IF can compromise sensitivity, once a shorter 2f_IF means image frequencies closer to the fRF pretended channel, and so image reject can be harder to implement. The receiver

sensitivity decreases for a low f_IF once higher power levels from the image can reach the mixer lowering SNR. Figure 3.5 depicts the trade off between choosing a high or low IF side, with the same image reject filter.

Select the desired channel (f_RF) with a nearby interferer (f_n) is hard to achieve for a high centred fIF. In opposition to a low fIF, where narrower bandwidth filters are more effective,

allows to remove the unwanted interferer.

Both image reject and channel select filters should afford a good attenuation outside the band of interest and low loss inside the pass band to minimize signal deterioration.

There are different adaptations from the basic architecture described in 3.3, as dual IF conversion, but advantages of one over the other can be relative [35]

(50)

Figure 3.4: Image rejection and channel select in heterodyne receivers

Figure 3.5: Rejection of image and channel selection differences for a high or low IF [7]

3.1.3 Super-heterodyne receivers

Super-heterodyne or heterodyne receivers are often described as only one type. There is no intention here to evidence differences although for coherency with literature it is presented in figure 2.7 a super-heterodyne block description.

In fact this will be the architecture used to implement the spectrum analyser prototype. The RF source signal used to the development stage is a controlled function generator. For that reason considerations about signal acquisition with an antenna, variable attenuator to protect sensible parts as the mixer, and image reject filtering will not be take in account. As so, the block description is present in diagram 3.6.

It will be used a LNA to increase overall sensitivity. A filter bank at the output of the mixer will provide variable RBW. A single high gain amplifier is placed at the end of the IF stage recovering the signal power to levels where a RF power detector can sense it.

This architecture provides incoming RF conversion and with a passive fixed tune BPF it is possible to select from the output of the mixer a specific incoming frequency. An important

(51)

characteristic to mention is the tuning ability once LO oscillating frequency will be successively scanning the input range. This matches the needs of a spectrum analyser design, once it makes frequency sweeps to measure the existing power in each sweep step. A microcontroller is used as the control center holding the human interface while reading input user instructions and keeps the Voltage Controlled Oscillator (VCO) sweeping in frequency the input range. For each tune step it happens a DC voltage readout at the output of the power detector. The digital microcontroller will also drive a graphic LCD, under a swept tune method to perform power analysis over frequency.

Figure 3.6: Spectrum analyser prototype block diagram

Recalling the theory of operation for a spectrum analyser and with intuit to understand the logical set of functions instructed by the chosen architecture, figure 3.7 illustrates what is spectrally happening in a swept tune analysis. It represents the input RF spectrum and the LO sweep range to perform down conversion. The signal at the output of the mixer passes through an IF filtering stage and the respective output spectrum is also represented.

Considering the mixer conversion loss of 10 dB and another 10 dB of attenuation inside the filter pass band. In total the signal channel is attenuated 20 dB from the RF input to the IF output. Outside of the filter passband attenuation is roughly 60 dB, allowing the receiver to be insensitive to nearby interferers.

For a BPF with a center frequency of f c = 70M Hz, while pretending to scan the input range from 880 to 950 MHz the LO is set to perform a low side injection as consequence to the selected fIF = fRF − fLO = 70M Hz. This implies the LO oscillating frequency to keep the

relation f_RF − f_LO = 70M Hz. Evoking the problem of image, it should be mentioned that frequencies higher than 950 MHz should not reach the mixer with penalty of image frequency fall inside the passband.

Three snapshots were take to the output spectrum of the BPF while the LO sweeps the RF input range. The three moments take place when f_LO= {810, 840, 870} M Hz being each distinguished with the graph color.

Therefore is detected a power increase inside the 5 MHz pass band only for f_LO = {810, 840} M Hz. When fLO = 870M Hz there is no power over fRF = 940M Hz resulting in

a no power variation at the output of the filter.

This sweeping process is continuously done and kept by the microcontroller. Decisions on the sweep time can be adjusted as frequency range, controlling the LO tuning range.

(52)

Figure 3.7: LO sweep

3.2 Components functionality

As result of the architectures description all the functions are implemented with RF elec-tronic components. Reduced sizes and high integration technology brings to market many different IC styles as differences in performance specifications. Even though general behaviour of RF components can be described. To ensure the correct service of each pieces it is need to select them with a closer look to match design specifications. Concerns about physical dimension and package style will be take in next chapter.

In general, RF components acts as blocks functions with an input and a output. These blocks may be mentioned as networks and allows RF design projects to be schematized as it as been so far.

Here it will be described the performed function by each block and the effect they have on circuit.

3.2.1 Amplifiers

RF designed amplifiers act as gain blocks. They provide power amplification to the input signal [36]. There are distinct ways to express gain. The gain of an amplifier can be expressed in dB as follow equation 3.1. In this case, the gain factor G is added to the input signal power Pin expressing the output power as in equation 3.2. For example, if the gain block in figure

3.8 has a gain G = 10dB and the input signal a power of P_in = −20dBm, the amplifier fit out Pout= −10dBm.

It is usual to find RF amplifiers already matched to be stable in 50 Ω systems. Amplifiers can be characterized by maximum power gain, noise figure, VSWR, DC power consumption

(53)

and the emitted power at harmonics distortion. Operation bandwidth is also preponderant for the RF project. The importance of these figures of merit change in order to the finality which is intended the amplifier.

Another important figure of merit is the Power Added Efficiency (PAE). This parameter demonstrate the efficiency of DC to RF power conversion. How smaller is the result of equation 3.3, higher is the DC power consumption. This implies with batteries life time and in case of base stations turns into an overeating of the device [36].

G = 10 log Pout Pin (3.1) Pout= Pin+ G (3.2) P AE = Pout− Pin PDC (3.3)

Figure 3.8: Gain block

Low noise amplifier

As the name implies, this is an amplifier with a low noise figure. These devices are usually used in the front of thed RF receiver chain determining the lowest signal possible to examine. This block in the spectrum analyser design is responsible for one of the figures of merit, sensibility.

Power amplifiers

General propose amplifiers, with higher power gain are used in the rest of the spectrum analyser schematic, but with the disadvantage of higher noise figure. This has no major consequences because the SNR levels are all ready high enough, and does not affect the power signal detection.

3.2.2 Mixer

Mixers are non linear designed devices that provide frequency translation. They can be active or passive and mixer’s ports are described in block diagram 3.9. Theoretically signals phase and amplitude are not disturbed, making it available to work with modulated signals [36, 37, 38, 39].

The fundamental operation of this devices, the frequency conversion, is obtained as the sum or difference between the two input signal, generating new ones.

(54)

Figure 3.9: Mixer ports diagram

This device can transform RF to a lower IF frequency easy to process in receivers. This operation is usually designated as down-conversion. Additionally it may reciprocal convert IF as a base band signal to a higher RF frequency, for efficient wireless transmission. This method is described as up-conversion. For this prototype project which is a radio receiver, the mixer will be used as a down-converter.

The non linear mixer designed response creates a group of output signals containing mul-tiplies of the input signals, with sums and differences all together. It is important to refer on this stage the importance of properly filtering the mixer output in order to remove the unwanted signals.

Considering v_RF, v_LO and v_IF as the signals presents at mixer’s respective terminals. The signals vRF and vLO can be described as:

vRF(t) = aRF(t) cos(ωRFt) (3.4)

vLO(t) = aLO(t) cos(ωLOt) (3.5)

Where ω is the angular frequency denoted as in 3.6 and a(t) represents the sinusoid am-plitude.

ω = 2πf (3.6)

Ideal the signal v_IF which is the mixer’s output, is described as a multiplication of the other two input signals.

vIF(t) = vRF(t)vLO(t) (3.7)

From equation 3.7 it cames: vIF(t) = aRF(t) 2 cos(ωRFt + ωLOt) + aLO(t) 2 cos((ωRFt − ωLOt) (3.8) = aRF(t)aLO(t) 2 (cos((ωRF+ ωLO)t) + cos((ωRF − ωLO)t) (3.9) Equation 3.9 reveals the sum and subtraction frequency terms. These are the second order responses [37].

ωRF + ωLO (3.10)

(55)

Considering equation 3.6 where is defined the angular frequency it is possible to express the mixer sum and difference output frequency terms as mentioned in equation 3.12

fIF = fRF ± fLO (3.12)

An important observation about equation 3.12 is it can be expressed as well in the way around:

−fIF = fLO± fRF (3.13)

In real world there are no difference in positive or negative frequencies. This is often miss explained in literature and can bring severe unwanted consequences to the system project if not considered.

For example, projecting the mixer to down-convert fRF for fIF = 70M Hz3. At mixer IF

output port it will appear a 70 MHz signal whenever f_RF − f_LO = 70M Hz or fLO− fRF =

−70M Hz.

So in practice the output interest terms to know about mixers are expressed as:

|f_RF ± f_LO| (3.14)

Therefore the receiver bandwidth is limited to the chose of the IF frequency, RF and LO range [3]. To note that mixers have RF and LO leakages to the IF output.

Another important parameters to know about a mixer on a practical design are the VSWR, conversion loss, ports isolation, LO power requirements, IP3 and noise factor.

Over the frequency translation operated by the mixer, it happens a power loss from the RF input signal to the IF output signal. This power loss is designated as conversion loss and mathematically expressed in equation 3.15.

Lc= 10 log

PRF

PIF

(3.15) In opposition to a device gain expressed in equation 3.1, passive mixers attenuate the signal and L_c is often expressed as a negative quantity in dB.

3.2.3 Voltage controlled oscillator

The primary function of an VCO is to produce a oscillating electric signal at a specific frequency. Furthermore the signal frequency is variable and voltage controlled. These compo-nents are widely used in super-heterodyne architectures as the LO signal source. The figures of merit of a VCO are the tuning range, respective tuning sensitivity (MHz/V), frequency stability (PPM/◦C), phase noise and output power.

3.2.4 Filters

Filters allow to reject some frequency bands and select others of interest. Typically in RF projects, filters are implemented in microstrip, resonant cavities or Surface Acoustic Wave (SAW) devices. The most important characteristics on filters are the VSWR, pass-band pass-bandwidth, insertion loss inside pass-band and attenuation outside pass-band.

(56)

Designations as Low Pass Filter, Band Pass Filter and High Pass Filter tells the frequency response characteristic. Cut off frequencies inform where the filter attenuation level changes. Filter attenuation inside the pass band should be low as opposition to the outside band of interest where it must attenuate other frequencies.

For this prototype is considered a filter bank in the IF section. This selects the frequency component of interest from the mixer providing different Resolution Bandwidths to spectra analysis. Therefore all filters in this bank should be centred to the same frequency to allow a constant LO sweep.

Each filter with a different pass bandwidth is selected to operation by the means of two additional blocks. One of then may be a RF switch digital controlled and the other a power combiner.

Figure 3.10: IF filter bank

3.2.5 RF switch

This device is able to route high frequency analogue signals trough different channels paths. It is an active device which may be digital controlled being suitable for automated systems. This devices are quite useful in laboratories environment where a signal can be conducted to different equipment tests.

As this project foresee an automatized digital circuit control, it turn to be an appropriate toll to be used in the filter bank, selecting the signal passage for the desired filter.

RF switches are characterized with VSWR, insertion loss, isolation between channels, P1dB and for exigent applications aspects as switching and settle times may also be detailed.

3.2.6 Power combiner

Power combiners are passive devices that can divide a single RF channel into different outputs or join together diverse transmission paths into one according to the propagation direction. It got a pre defined insertion loss for the frequency band which is designed. For the filter bank in this project is necessary to conduct the output of three filters to a single power detector, and this component turns to be serviceable, joining the signal paths.

(57)

The most important characteristics on a combiner are VSWR, insertion loss and ports isolation [36].

3.2.7 Power detector

This component is the last one in the receiver chain and it will be where the incoming RF signal is finally delivered. After the IF stage the signal’s frequency will be centred at 70MHz, so the power detector needs to work in the same frequency range.

These devices are design for RF applications. In order to the input power, it sets an output DC voltage. Knowing the typical output voltage curves it is possible to measure the RF power in dBm.

3.2.8 Digital control - Microcontrollers

Behind the super-heterodyne architecture theory of operation it is need to control some variables in the RF system, to validate measurements and turn them available for users. By other means, translate the electric signals to human comprehension domain. For electronic projects, microcontrollers may be a very attractive option with reduced sizes, affordable prices and ease to use interfaces.

Microcontrollers likewise computers on chips have a processor unit provided with a va-riety of embedded peripherals such as non volatile flash memories, Electrically Erasable Programmable Read Only Memory (EEPROM), volatile memories as Random Access Mem-ory (RAM) Static Random Access MemMem-ory (SRAM), input and output communication de-vices as Universal Asynchronous Receiver Transmitter (UART), Universal Serial Bus (USB) or even Ethernet. Alongside with ADCs, Input Output (IO) digital pins and Digital to Analogue Converters (DACs) these devices are able to sense and interact with external components and handle automated processes.

The programmed tasks can be done with a high-level abstraction4 allowing to the code development with a programming language as C, avoiding the intrinsic native languages of each manufacturer.

In the architecture variables that require control are the electric DC voltage applied on the VCO, the channel selection in the IF filter bank and the incoming power interpretation. With an electric visual screen it is possible to bring electromagnetic graphics representations to sight.

4

High-level programming languages provides abstraction to the minutiae details of the processor hardware while low-level languages turn to be keen in those specifications

(58)

Handled RF spectrum analyser using Arduino

Vitor Miguel

Melo Soares

Analisador de espectros portátil para RF usando

Arduino

Vitor Miguel

Melo Soares

Analisador de espectros portátil para RF usando

Arduino

Handled RF spectrum analyser using Arduino

Contents

List of Figures

List of Tables

Acronyms

Chapter 1

Introduction

1.1

Motivation

1.2

Objectives

1.3

Dissertation outline

Chapter 2

Background theory

2.1

Signals

2.2

Frequency spectrum

2.3

Time and frequency domain

2.4

Spectrum analyser theory of operation

2.5

Spectrum analyser characteristics

2.6

RF design basic concepts

Chapter 3

Architectures

3.1

Receiver architectures

3.2

Components functionality